Scalable Data Ablation Approximations for Language Models through Modular Training and Merging
Paper
•
2410.15661
•
Published
Datasets and models for EMNLP paper "Scalable Data Ablation Approximations for Language Models through Modular Training and Merging"